JPH11312073A - System for recognizing meaning - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a meaning recognizing system for recognizing meaning, and communicating the voice input instruction of user's language as an instruction independent of language to an application(AP), and for loading the AP, and constructing a user I/F capable of voice input in user's language. SOLUTION: A voice signal from a voice input device 10 is recognized by a voice recognizing part 30, and converted into a meaning number independent of language by a recognition word and meaning number converting part 40, and transferred to an AP 50. The meaning number is included in one part of the processing content in the AP 50, and the processing is executed, based on the meaning number. Also, in loading the AP 50, the meaning number is obtained, and a registered word corresponding to the meaning number used by the AP 50 is obtained from an outside description file so that a meaning number and registered word list 41 can be generated, and the recognition word and meaning number converting part 40 is updated. The registered word is communicated to a voice recognizing part 30, and a voice recognition engine 31 is tuned up so that a customized user interface can be constructed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a system for providing an interface for an application capable of voice input, and a system for recognizing the meaning of a user's input instruction to the application by voice and notifying the application. .

[0002]

2. Description of the Related Art In recent years, the advancement of computer systems and the development of multimedia processing technology have made it possible to provide advanced human-machine interfaces, and an application development environment that allows direct interaction with humans is being prepared.

An application using the speech recognition technology in the prior art is roughly divided into a speech input unit, a speech recognition unit, and an application program. The voice input unit converts a real voice into a voice signal via a voice input device such as a microphone. Alternatively, data may be read from a file in which an audio signal is recorded. The voice recognition unit recognizes a word represented by the input voice signal by template matching with a registered voice signal pattern, and notifies the application program of the recognized word. The application program executes subsequent processing using the recognized word notified by the voice recognition unit.

The features of the above-mentioned conventional speech recognition application are as follows. The first feature is that the application program has a one-to-one correspondence
Is described as a processing program corresponding to. In other words, the speech recognition unit notifies the application program of the result of matching from the registered recognition target words, so that the application internally classifies the subsequent processing into necessary processing in advance for each recognition target word and individually describes the processing. Need to be kept.

A second feature is that the application depends on the language of the word to be registered and recognized. In other words, since words other than the recognition target words registered in the speech recognition unit cannot be recognized, the application program is described based on the registered language, and is language-dependent.

[0006]

At present, the development of various applications is accompanied by the sophistication and complexity of the applications to be developed, the diversification of changes in market needs, the internationalization of products, and the like. Shorten development time,
Multilingual support is required. However, the conventional speech recognition application introduced above has the following problems.

[0007] Regarding the first feature of the conventional speech recognition application, in which an application application is described for each speech recognition target word, tuning for increasing the number of development steps, changing / adding the recognition target word, and the like. There was a problem of difficulty. In other words, it is necessary to classify the processing content into necessary processes for each recognition target word in advance and describe them separately.
This affected all relevant parts of the application program, and the description had to be changed.

The second feature of the conventional speech recognition application is that the application depends on the language of the word to be registered / recognized, which causes problems such as an increase in development man-hours and difficulty in coping with multiple languages. was there. In other words, there is a language-dependent part inside the application,
Changing the word expression in accordance with the system user's language, dialect, etc. would affect all language-dependent parts, and required localization work to change the description.

In a conventional speech recognition application, it is possible to update and add a word to be recognized by updating and adding a word table to be recognized. All of them are affected, and according to the method of replacing the recognition target word table for the support of other languages, the application program itself is eventually changed to the language of the replaced recognition target word table due to language dependency. It was necessary to develop and replace a localized version to correspond.

The present invention has been made in view of the above-mentioned problems of the conventional speech recognition system, and makes the relationship between the language-dependent speech recognition unit and the application mutually independent, so that a language most suitable for the user as a user interface can be used. The purpose of this system is to provide an environment for using dialects, and to make it easy for speech recognition-related applications to handle multiple languages by providing a system that focuses on language-independent semantic recognition on the application side. And It is another object of the present invention to provide a meaning recognition system in which updating and changing of a recognition target word table is easy and has minimal influence on an application program description portion of an application.

[0011]

In order to solve the above-mentioned problems, a semantic recognition system according to the present invention comprises a voice signal input unit for receiving a voice signal to be input, and a corresponding word based on the input voice signal. A speech recognition unit for recognizing, a recognition word / meaning number conversion unit having a meaning number representing the meaning of the word and a meaning number / registration word list which is a list of registered words belonging to the meaning number, an application interface, and a meaning An application that handles a number as data is provided, the corresponding word is recognized by the voice recognition unit based on the voice signal input to the voice signal input unit, and the recognized word is supported by the recognized word / semantic number conversion unit. And converts the converted meaning number to the application via the application interface. And wherein the door.

With this configuration, it is possible to provide a user interface in the most convenient language and dialect for the user, and to convert a registered word of the language-dependent speech recognition unit into a language-independent meaning number. The application can be notified and the application can be described as language independent.

Next, the meaning recognition system preferably includes a meaning number / registration word list updating unit for updating the meaning number / registration word list of the recognition word / meaning number conversion unit.

With this configuration, the meaning recognition system can be tuned up by adding / changing the meaning number / registered word list. Next, the semantic recognition system records a word / semantic system table that systematically describes the semantic numbers that systematically classify the meanings and codes each semantic unit, and the registered words belonging to each semantic number. It is preferable to provide the external description file described above.

With this configuration, a word / semantic system table that systematically represents the relationship between a word and a semantic unit corresponding to the word can be configured to be separable from the application main body, and the user's usage environment By exchanging external description files according to the above, basic data for localizing an application can be prepared.

Next, the application is an application that includes a meaning number in at least a part of processing data, and the recognition word / meaning number conversion unit is included in the application via the application interface. A meaning number acquiring unit for acquiring a meaning number, a meaning number for extracting a registered word corresponding to the acquired meaning number from the word / semantic system table of the external description file, and generating the meaning number / registered word list.
Preferably, a registered word list generation unit is provided.

With this configuration, an application is described based on a language-independent semantic number, and when localizing the semantic recognition system, words in a local language corresponding to the semantic number used in the application are automatically extracted. And the number of man-hours for localization work can be significantly reduced.

Next, the application is an application that includes a meaning number in at least a part of the processing data, and is based on the meaning number included in the application itself. And a processing step of notifying the meaning number and the registered word to the recognition word / meaning number conversion unit, wherein the recognition word / meaning number conversion unit obtains the corresponding registration word from the application. It is preferable to include a meaning number / registered word list generation unit that generates the meaning number / registered word list based on the obtained meaning number and the registered word.

With this configuration, an application can be described with a language-independent semantic number, and in localizing the semantic recognition system, a word in a local language corresponding to the semantic number used by the application can be notified. The man-hour for localization work can be greatly reduced.

Next, the recognition word / meaning number conversion unit:
A registered word notifying unit for notifying the voice recognition unit of a registered word listed in the generated meaning number / registered word list, wherein the voice recognition unit includes a voice recognition engine based on the notified registered word. It is preferable to include a speech recognition word updating unit that updates the content.

With this configuration, it is possible to specify a word that needs to be recognized by the voice recognition unit for use in an application and automatically generate the content of the voice recognition engine required for the word recognition. Reduction is possible.

Next, the semantic recognition system includes a language specification unit for specifying a language used by a user, an external description file for each of a plurality of languages as the external description file, and a voice recognition engine of the voice recognition unit. And a speech recognition engine for each of a plurality of languages, and the meaning number / registered word list generating unit generates a meaning number / registered word list based on an external description file of the language specified by the working language specifying unit. The voice recognition unit uses a voice recognition engine of a language specified by the language-in-use specification unit, and provides a multilingual user interface capable of instructing a user to input the application in a plurality of languages. Is preferred.

With this configuration, a customizing operation can be easily performed in accordance with the user's language and expression. Also, by specifying an external description file in multiple languages and a speech recognition engine, it is possible to automatically generate a list of meaning numbers and registered words in multiple languages and update the contents of the speech recognition engine. Multilingual support that allows input instructions can be provided. Here, the language may include a dialect, and in addition to languages such as Japanese and English, may include dialects of various languages such as the Kansai dialect and the Tohoku dialect.

Next, the semantic recognition system includes an externally described file server that provides the externally described file, an application server that provides the application, the voice signal input unit, the voice recognition unit, and the recognized word / meaning. Preferably, a semantic recognition system client including a number conversion unit and the application interface is provided, and the externally described file server, the application server, and the semantic recognition system client are connected via a computer network.

With this configuration, it is possible to construct an externally described file in a client server configuration as an externally described file server arranged on a network, and it is necessary to prepare an externally described file for each language separately for the semantic recognition system. Disappears.

Next, a computer-readable recording medium on which the semantic recognition processing program according to the present invention is recorded has an audio signal input processing step of receiving an input audio signal, and a computer-readable recording medium corresponding to the input processed audio signal. Voice recognition processing step of recognizing a word to be performed; a meaning number representing a meaning and a meaning number / registered word list generating step of generating a meaning number / registered word list as a list of registered words belonging to the meaning number; A recognition word / meaning number conversion processing step of converting the word recognized by the processing step into a corresponding meaning number based on the generated meaning number / registered word list, and a meaning number notifying process of notifying the converted meaning number to the application And recording a semantic recognition processing program having steps.

With this configuration, it is possible to provide a semantic recognition system in which a user can use the most convenient language and dialect as a user interface by using a computer system, and furthermore, a registered word of a language-dependent speech recognition unit. Can be converted to a language-independent meaning number and notified to the application, and the application can be described as language-independent.

Next, it is preferable that the semantic recognition processing program includes a meaning number / registered word list updating processing step of updating the contents of the meaning number / registered word list.

With this configuration, it is possible to perform processing for adding / changing the meaning number / registered word list, and to tune up the processing by the meaning recognition system. Next, the semantic recognition processing program generates a word / semantic system table that systematically describes the semantic numbers in which the semantic units are coded by systematically classifying the semantic units and the registered words belonging to the respective semantic numbers. It is preferable to have a recorded external description file.

With this configuration, in the semantic recognition processing, a word / semantic system table that systematically represents the relationship between a word and a semantic unit corresponding to the word can be configured to be separable from the application main body. By exchanging external description files according to the usage environment of the user, basic data for localizing an application can be prepared.

Next, the semantic recognition processing program includes a semantic number acquiring step of acquiring the semantic number included in the application, wherein the application is an application including a semantic number in at least a part of processing data; Wherein the meaning number / registered word list generation processing step extracts the registered word corresponding to the acquired meaning number from the word / semantic system table of the external description file to generate the meaning number / registered word list And a registered word notification processing step of notifying the registered word listed in the generated meaning number / registered word list to the voice recognition processing step, wherein the voice recognition processing step includes the notified registered word. Speech recognition word update processing step of updating the content of the speech recognition engine based on the It is preferably provided.

With this configuration, an application is described with a language-independent semantic number, and a local language word corresponding to the semantic number used in the application is automatically extracted when localizing the semantic recognition processing step. And the man-hour for localization work can be significantly reduced.

Next, the semantic recognition processing program includes a use language specification processing step for specifying a language used by a user, an external description file for each of a plurality of languages as the external description file, and a speech recognition processing step. A speech recognition engine for each of a plurality of languages is used as a speech recognition engine to be used. In the meaning number / registered word list generation processing step, the meaning number is determined based on an external description file of the language specified in the use language designation processing step. A registered word list is generated, and in the voice recognition processing step, an instruction to input the application in a plurality of languages can be given to the user by using a voice recognition engine of a language specified by the language specification processing step. Provide a multilingual user interface. It is preferred.

With this configuration, a customizing operation can be easily performed according to the language and expression used by the user of the semantic recognition system. Also, by specifying a plurality of external description files, it is possible to automatically generate a meaning number / registered word list in a plurality of languages and update the content of a speech recognition engine, and to instruct an application to input speech in a plurality of languages. It can be multilingual.

Next, the semantic recognition processing program provides a processing program for an externally described file server, which is provided on a network, including an externally described file providing processing step of providing the externally described file, and the application. A processing program for an application server provided on a network having an application providing processing step, a processing program for the externally described file server, and a semantic recognition system client for communicating over the network with the processing program for the application server It is preferable that the program is provided with a processing program.

With this configuration, it is possible to construct a semantic recognition system having a client-server configuration in which an externally described file and an application are provided from a server arranged on a network.

[0037]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a semantic recognition system according to an embodiment of the present invention will be described with reference to the drawings.

(Embodiment 1) The semantic recognition system according to Embodiment 1 provides a user interface capable of voice input to a user, and the original content of each word corresponding to the input content of the user to an application is provided. This is a system that provides an application interface that communicates as language-independent by performing semantic recognition by focusing on the meaning of. In the first embodiment, the application is a game including a process of controlling the movement of an object appearing on the screen by voice instructions, the user is a person who uses Japanese, and a Japanese environment is constructed as a user interface. I will.

An outline of the overall configuration of the semantic recognition system according to the first embodiment and an overall image of the processing flow of the system will be described with reference to the drawings. FIG. 1 is a schematic configuration diagram of the meaning recognition system according to the first embodiment. FIG.
Represents the overall image of the processing flow of the present system as processing steps.

As shown in FIG. 1, the semantic recognition system according to the first embodiment is roughly divided into a voice input device 1
0, a voice signal file 20, a voice recognition unit 30, a recognized word / meaning number conversion unit 40, and an application 50. Although not shown, a controller, a memory, and the like necessary for control processing of the entire system are provided.

The voice input device 10 may be any device that converts an input voice into an electric signal, such as a microphone or a telephone. The audio signal file 20 is a file in which input audio is edited and recorded in advance. Voice recognition unit 3
0 includes the speech recognition engine 31, and the speech recognition engine 31 can be switched according to the language used by the user. The recognized word / meaning number conversion unit 40 includes a meaning number / registered word list 41 indicating a correspondence between a word recognized by speech recognition and a meaning number indicating the meaning of the word,
A meaning number / registered word list changing unit 42 for adding / changing the contents of the registered word list 41 is provided. Here, the semantic number codes a semantic unit, which is a set of words having the same meaning, which systematically classifies the meaning of the words. The semantic number is an intermediate code given to the original semantic content of each word that does not depend on the language, and is introduced by paying attention to the fact that the meaning intended by the language does not change even if the language is different Things. This semantic number may cover words belonging to the language system, but in consideration of system resources and effectiveness,
It may cover words that can be used in a field that is expected to be used as an application such as a game.

An application interface is provided between the recognition word / meaning number conversion unit 40 and the application 50 for transmitting the voice input contents of the user.
The application 50 includes a semantic number as a part of the processing content.

The overall image of the processing flow of the semantic recognition system is as follows. First, the user inputs, in his / her own words, an instruction to be input to the application 50 via the voice input device 10 such as a microphone provided as a user interface. Here, "stop" is input in Japanese as an example. Next, the input “stop” voice is converted into an electric signal and input to the voice recognition unit 30. In addition, instead of the voice input by the voice input device 10, the words to be input in advance can be edited / edited.
A corresponding audio signal may be input from the recorded audio signal file 20 to the audio recognition unit 30 (step S10 in FIG. 2).
1).

Next, the speech recognition section 30 recognizes the given speech signal. The voice recognition engine 30 includes a voice recognition engine 3
In this example, it is assumed that a Japanese speech recognition engine is mounted. The speech recognition unit 30 executes processing such as morphological processing and template matching processing, which are normal speech recognition processing, and recognizes an input word. Here, the voice of "stop" is recognized as word data of "Tomare". (Step S10 in FIG. 2)
2).

Next, the recognized word data recognized in step S102 is converted into a corresponding meaning number by the recognized word / meaning number conversion unit 40. The recognized word / meaning number conversion unit 40 refers to the meaning number / registered word list 41 and searches for a meaning number corresponding to the recognized word data. Here, the recognition word data “Tomare” is converted into the meaning number “1” (Step S103 in FIG. 2).

Here, examples of the meaning number / registered word list are shown in FIGS. The list shown in FIG. 3 is a structure having, as element members, “symbol number”, “semantic number”, and “registered word”, and “category” and “word explanation” as word accessory information as element members. As an array. The symbol number is a number uniquely assigned to each entry, the registered word is a character string to be recognized, the category is a coded classification of the registered word, and the description for the word is , A description for a registered word. When no explanation is given to a category and a word, for example, “NULL” is assigned.

In the list shown in FIG. 4, in addition to the element members shown in FIG. 3, as an element member, an element member for managing a linked list “address of the next table” is added.
The structure is such that individual entries can be easily inserted and deleted.

Next, the meaning number obtained in step S103 is given to the application 50 via the application interface (step S104 in FIG. 2).
The application 50 includes a meaning number in a part of the processing content, and the processing corresponding to the input meaning number is executed. The application 50 is a game including a process of controlling the movement of an object appearing on the screen by a voice instruction. In this example, the process of stopping the movement of the object is performed for the meaning number “1” (step in FIG. 2). S105). In other words, in the example given here,
In response to a voice instruction from the microphone saying “stop” by the user, a process of stopping the movement of the object appearing on the screen is performed.

This meaning recognition process (step S101 in FIG. 2)
Steps S105 to S105) are repeated as appropriate in accordance with the progress of the game by the user. The above is the first embodiment.
1 shows an outline of a configuration of a meaning recognition system according to the first embodiment and an overall image of a processing flow by the present system.

As described above, the instruction content can be transmitted to the application 50 by the voice input from the input device 10. The semantic recognition system according to the first embodiment can provide a user interface for voice input to a user, and can provide an interface for an application that conveys language-independent semantic contents to an application.

In the first embodiment, the processing in Japanese is given as an example. However, it goes without saying that the system can be adapted to other languages by tune-up of the system. That is, the speech recognition engine 31 of the speech recognition unit 30 is replaced with the speech recognition engine of the language to be used, and the meaning number / registered word list 4
This can be dealt with by replacing 1 with the meaning number / registered word list of the language you want to use. Here, the meaning numbers of the meaning number / registered word list are language-independent because the same meaning number is assigned to the same meaning content even if the language used is different, and the application 50 has the same meaning number. The same meaning number is transmitted to the voice input instruction of the meaning content, and the application 50 executes the same processing for the same meaning number. As described above, it is possible to provide an application interface that conveys the voice input content of the user to the application as language-independent by performing meaning recognition by focusing on the original meaning of each word.

The meaning number / registration word list changing unit 42 of the recognition word / meaning number conversion unit 40 causes the meaning number
The contents of the registered word list 41 can be added or changed. For example, as described in the description of the list shown in FIG. 4, by using the element members of "address of next table", entry insertion and deletion can be easily performed as shown in FIG. FIG. 5 shows that in the meaning number / registered word list 41, the recognition word “uehetob” and the meaning number “1”
In this example, a set of a recognition word and a meaning number consisting of 5 ″ is added as the 101st entry. By this meaning number / registration word list changing unit, the meaning recognition system can be tuned up according to the use environment. It is possible.

(Second Embodiment) A semantic recognition system according to a second embodiment will be described with reference to the drawings. The second embodiment is a system that provides a user interface by voice input to the user and an application interface for transmitting the user's voice input to the application as language-independent, as in the first embodiment. A feature of the semantic recognition system according to mode 2 is that the application is loaded and information is obtained from the application, and a user interface for operating the application is customized and constructed.

In the second embodiment, as in the first embodiment, the application is a game including a process of controlling the movement of an object appearing on the screen by voice instruction. As an example, an English environment is constructed.

An outline of the overall configuration of the semantic recognition system according to the second embodiment and an overall image of the processing flow of this system will be described with reference to the drawings. FIG. 6 is a schematic configuration diagram of the meaning recognition system according to the second embodiment. FIG.
Represents the overall image of the processing flow of the present system as processing steps.

As shown in FIG. 6, the semantic recognition system according to the second embodiment is roughly divided into
0, voice signal file 20 (not shown), voice recognition unit 3
0, an external description file 60 in addition to the recognition word / meaning number conversion unit 40 and the application 50. Although not shown, a controller, a memory, and the like necessary for control processing of the entire system are provided. Elements similar to those described in the first embodiment are assigned the same numbers.

The application 50 is an application in which at least a part of the processing data includes a meaning number. The external description file 60 is an external file separated from the application itself, and may be stored in an accessible secondary storage medium such as a hard disk or a CD-ROM, and need not be integrated with the program. The content of the file recorded in the external description file 60 is a table that systematically describes the meaning numbers that systematically classify the meanings and encodes each meaning unit and the registered words belonging to each meaning number. is there. This word / semantic system table may include a registered word group covering the words belonging to the language system and a semantic number group corresponding to the registered word group. It may cover words that can be used in the field where the use is planned.

FIG. 8 shows an example of the external description file.
“State-name” is an attribute that specifies a name for describing a category when a word is registered, and “Language-I”
D "is an attribute representing the language of the word.
D, sub-language ID, and the like. “Dialect” is an attribute related to the dialect of the language, and the default is a standard language (Standard). “VCMD” is a voice command (Voice c
ommand), which is a set of a meaning number and a registered word which are each entry of the meaning number / registered word list 41. “#” Indicates that the line is a comment.

The recognized word / meaning number conversion unit 40 includes a meaning number acquiring unit 43, a meaning number / registered word list generating unit 44, and a registered word notifying unit 45. Meaning number acquisition unit 43
Is to acquire the meaning number described in the application 50 via the application interface. The meaning number / registered word list generation unit 44 extracts all registered words corresponding to the meaning number obtained from the external description file 60 and generates the meaning number / registered word list 41. The registered word notification unit 45 outputs the recognized word
The registered words listed in the meaning number / registered word list 41 generated by the meaning number conversion unit 40 are notified to the speech recognition unit 30.

The speech recognition unit 30 includes a speech recognition engine 31 and a speech recognition word updating unit 32 corresponding to the language used by the user.
And The speech recognition word update unit 32 updates the content of the speech recognition engine 31 based on the registered word notified from the registered word notification unit 45.

The overall processing flow of the semantic recognition system according to the second embodiment is as follows. First, the application 50 is loaded into the semantic recognition system. Loading is performed by an IO device such as a CD-ROM drive, download from a network (not shown), or the like (step S201 in FIG. 7). Here, as an example, it is assumed that the loaded application 50 is a game with movement control of an object appearing on the screen, and a description based on a meaning number of an object movement instruction is included in a part of the processing program. Shall be.

Next, the meaning number acquiring unit 43 of the recognized word / meaning number converting unit 40 acquires the meaning number described in the application 50 via the application interface (step S202 in FIG. 7). Here, it is assumed that the meaning number “1” has been acquired, for example.

The meaning number obtained in step S202 is passed to the meaning number / registered word list generation unit 44. The meaning number / registered word list generation unit 44, for example, as shown in FIG. 9, expands the acquired meaning number into the meaning number / registered word list 41 and holds it (step S20 in FIG. 7).
3).

Next, the meaning number / registered word list generation unit 4
4 extracts the registered word corresponding to the meaning number in the meaning number / registered word list 41 with reference to the word / semantic system table of the external description file 60 to generate the meaning number / registered word list 41 (FIG. 7 steps S204). An example of the generated meaning number / registered word list 41 is shown in FIG. As described above, the instruction group used in the application is generated as the instruction group in the language used by the user from the intermediate code having the meaning number. Here, an instruction group in English is generated. As can be seen from FIG. 10, all registered words in the external description file 60 corresponding to the respective meaning numbers are extracted from the respective meaning numbers, and are expanded into a list. For example, the meaning number “1” representing the meaning of “stop”
In addition to "STOP", "HALT" and other English words with a direct meaning of "stop", "STAY", "DON'T MOV"
A synonym such as "E", which gives a similar action to the object, may also be used.

Next, the registered word notifying unit 45 of the recognized word / meaning number conversion unit 40 notifies the speech recognition unit 30 of the registered words listed in the meaning number / registered word list 41 generated in step S204 (FIG. 7 Step S205). Here, registered words such as “STOP” and “HALT” shown in FIG. 10 are notified.

Next, the voice recognition unit 30 proceeds to step S20.
Based on the registered word notified in step 5, the content of the speech recognition engine 31 is updated by the speech recognition word updating unit 32. That is, the content of the recognition pattern of the voice signal used by the voice recognition engine is updated, and a user interface capable of voice recognition of the registered word notified in step S205 in the language used by the user is constructed (step S20 in FIG. 7).
6).

The above steps S201 to S20
By executing the processing in step 6, the user input instructions used in the loaded application are automatically converted from the semantic numbers, which are language-independent intermediate codes, to the instructions in the language used by the user, and the A user interface based on input can be constructed.

The above steps S201 to S20
After the user interface customization processing by the user 6, the user can input his / her own language by voice and enjoy applications such as games by executing the processing steps of FIG. 2 described in the first embodiment. Further, since the meaning number / registered word list 41 used in the application is generated according to the application, the amount of use of the main memory can be optimized and reduced. In addition, since the application can describe the user input command in a language-independent intermediate code centered on the semantic number, the man-hour for localizing the application developer to another language can be eliminated.

In the above description, an example of constructing a voice input user interface in English has been given as an example. However, even in other languages, the basic configuration of the semantic recognition system may be changed according to the environment in which the semantic recognition system is to be used. It goes without saying that a user interface can be similarly constructed by providing a basic external description file and a speech recognition engine for each language. Further, in the above description, the recognition word / meaning number conversion unit obtains a meaning number described in the application,
The registered word list generating unit is provided, but instead of the semantic number acquiring unit and the semantic number / registered word list generating unit, the application uses the words of the externally described file based on the semantic numbers described in the application itself. A processing step of extracting a corresponding registered word from the semantic system table; and a processing step of notifying the meaning number and the registered word to the recognition word / meaning number conversion unit. A configuration may be provided that includes a meaning number / registered word list generation unit that generates a meaning number / registered word list based on the given meaning number and registered word.

(Embodiment 3) A semantic recognition system according to Embodiment 3 will be described with reference to the drawings. In the third embodiment, as in the first and second embodiments, a user interface by voice input to the user is provided.
This is a system that provides an application interface that conveys the user's voice input to the application as language-independent. The feature of the semantic recognition system according to the third embodiment is that the target language of the user interface is specified by the user. It can be freely selected and updated, and can handle multiple languages simultaneously. As an assumed usage environment of the semantic recognition system,
It is suitable when the user uses a plurality of languages or when there is a possibility that a plurality of people who speak different languages may use the languages at the same time. In the third embodiment, particularly, the construction of a user interface capable of simultaneously handling a plurality of languages will be described.

In the third embodiment, similar to the first embodiment, the application is a game including a process of controlling the movement of an object appearing on the screen by voice instruction. Assuming users, an example is given in which a Japanese input environment and an English input environment are constructed as user interfaces.

An outline of the entire configuration of the meaning recognition system according to the third embodiment and an overall image of the processing flow by the present system will be described with reference to the drawings. FIG. 11 is a schematic configuration diagram of a semantic recognition system according to the third embodiment. FIG. 12 shows an overall image of the processing flow of the present system as processing steps.

FIG. 11 has a configuration substantially similar to that of FIG.
The semantic recognition system according to the third embodiment includes a used language specifying unit 70, and a user specifies a desired language. The external description file 60 is prepared for each language, and includes a Japanese external description file 60a, an English external description file 60b, and a German external description file 60.
c etc. are provided. Also, a speech recognition engine 31 of the speech recognition unit 30 is prepared for each language, and a Japanese speech recognition engine 31a and an English speech recognition engine 31 are provided.
b, a German speech recognition engine 31c and the like. The same elements as those in FIGS. 1 and 6 are assigned the same numbers, and description thereof will be omitted. The arrows between the elements in FIG. 11 focus on the exchange of language-designated data and the like in processing step S301 to be described later for convenience, and others are omitted.

The overall processing flow of the semantic recognition system according to the third embodiment is as follows. The description that overlaps with the portion described with reference to FIG. 7 and the like in the second embodiment will be briefly described. First, the user specifies a language to be used by the used language specifying unit 70 (step S301 in FIG. 12). Here, two languages, Japanese and English, are designated. The language information specified and input by the used language specifying unit 70 is notified to the registered word / meaning number conversion unit 40 and the speech recognition unit 30 as language ID information. Here, a Japanese external description file 60a and an English external description file 60b are prepared,
A Japanese speech recognition engine 31a and an English speech recognition engine 31b are prepared.

Next, the application 50 is loaded into the semantic recognition system (step S302 in FIG. 12). Next, the meaning number acquisition unit 4 of the recognition word / meaning number conversion unit 40
3 acquires the meaning number described in the application 50 via the application interface (step S303 in FIG. 12). Here, for example, the meaning number “1” is acquired.

The meaning number obtained in step S303 is passed to the meaning number / registered word list generation unit 44. Similar to the second embodiment, the meaning number / registered word list generation unit 44 expands the acquired meaning number into the meaning number / registered word list 41 and holds it, as shown in FIG. 9 (step S304 in FIG. 12).

Next, the meaning number / registered word list generation unit 4
No. 4 refers to the external description file specified based on the language ID notified in step S301. Here, the Japanese external description file 60a and the English external description file 60b are referred to. Referring to the word / semantic system table of each external description file 60, the registered words corresponding to the meaning numbers in the meaning number / registered word list 41 are extracted to generate the meaning number / registered word list 41 (step in FIG. 12). S305). An example of the generated meaning number / registered word list 41 is shown in FIG. As described above, the instruction group used in the application is generated as an instruction group in a plurality of languages specified by the user from the intermediate code having the meaning number. Here, an instruction group in Japanese and an instruction group in English are generated.

Next, the registered word notifying section 45 of the recognized word / meaning number conversion section 40 notifies the speech recognition section 30 of the registered words listed in the meaning number / registered word list 41 generated in step S305 together with the language ID. (Step S306 in FIG. 12).

Next, the voice recognition unit 30 proceeds to step S30.
The content of the speech recognition engine 31 is updated by the speech recognition word updating unit 32 based on the registered word and the language ID notified in step 6. In other words, here, the Japanese speech recognition engine 3
1a and the English speech recognition engine 31b are tuned up (step S307 in FIG. 12).

Steps S301 to S30
7. By executing the processing in step 7, the user input instructions used in the loaded application are automatically converted into the instructions in a plurality of languages used by the user, and the multilingual user is input by voice input in a plurality of languages. An interface can be built.

After the user interface is customized by the above steps, the operation by voice input to the application 50 executes the same processing as the processing described in the first embodiment, and enjoys the application such as a game in a multilingual environment. it can.

In the above description, an example of constructing a voice input user interface in Japanese and English is given as an example, but it goes without saying that the present invention can be similarly applied to other languages. The language includes dialects, and is applicable to dialects of various languages such as Kansai dialect and Tohoku dialect in addition to languages such as Japanese and English.

(Embodiment 4) A semantic recognition system according to Embodiment 3 will be described with reference to the drawings. The third embodiment is an example in which the semantic recognition system shown in the first to third embodiments is constructed in a client-server configuration.

FIG. 14 shows the overall schematic configuration of the system. As shown in FIG. 14, an application server 80 and an externally described file server 90 are provided on a network.
Is arranged. Various applications 50 are registered in the application server 80, and the application 50 is provided by a request from the client 100. In the external description file server 90, an external description file 60 for each language and each dialect is registered, and the external description file 60 of the semantic recognition system shown in FIGS.
Can be remotely accessed on the network. That is, the semantic recognition system client 100
In response to a reference request with a meaning number from, a registered recognition word corresponding to the meaning number is notified.

Although not shown, the semantic recognition system client 100 has a voice input device 10, a voice signal file 20, a voice recognition unit 30, a voice recognition engine 31, a voice recognition word updating unit 32, a recognition word Word / meaning number conversion unit 40, meaning number / registered word list 4
1, a meaning number / registered word list changing unit 42, a meaning number acquiring unit 43, a meaning number / registered word list generating unit 44, a registered word notifying unit 45, and a language designation unit 70.

Although not shown, both the server and the client have a communication interface for network connection. The network 110 may be any network that can communicate data, and is a network such as a local area network or the Internet, and may be a dedicated line, a public line, a wired line, or a wireless line.

With the above configuration, it is possible to construct an externally described file in a client server configuration as an externally described file server arranged on a network, and to convert the externally described file for each language into a semantic recognition system. There is no need to prepare for.

(Embodiment 5) The semantic recognition system according to the present invention is described as a program having processing steps for realizing the semantic recognition system, and is provided by being recorded on a computer-readable recording medium and provided. Can be constructed. As shown in the example of the recording medium shown in FIG. 15, a recording medium storing a program having processing steps for realizing the meaning recognition system according to the present invention is a portable recording medium such as a CD-ROM 123 or a flexible disk 124. 122 as well as the recording medium 12 in the recording device on the network.
1 or a recording medium 126 such as a hard disk of a computer or a RAM. When the program is executed, the program is loaded on the computer 125 and executed on the main memory.

[0089]

According to the semantic recognition system of the present invention, the user is provided with a user interface for voice input in a language used by the user, and the user can input a voice or the like in a language of his / her own language for a game or the like. You can enjoy the application. In addition, it is possible to provide an application interface that conveys language-independent semantic contents to the application, and the application can describe user input commands with language-independent intermediate code centered on semantic numbers. Localization work for the user in another language can be eliminated.

Further, according to the semantic recognition system according to the present invention, since only the meaning number / registered word list used in the application is generated each time depending on the application, the capacity of the main memory used can be reduced. Speech recognition speed increases.

Further, according to the semantic recognition system of the present invention, a user input command group used in a loaded application is automatically converted into a command group in a plurality of languages used by the user, and a plurality of languages are used. A multilingual user interface by voice input can be constructed.

Further, in the semantic recognition system according to the present invention, by constructing the externally described file in a client server configuration as an externally described file server arranged on a network, the externally described file for each language is semantically recognized. There is no need to prepare individual systems,
It is possible to diversify the usage form and reduce the price of the system.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram of a semantic recognition system according to a first embodiment of the present invention;

FIG. 2 is a flowchart showing an overall processing flow of the semantic recognition system according to the first embodiment of the present invention;

[Figure 3] Example of meaning number / registered word list

Fig. 4 Example of meaning number / registered word list

[Fig. 5] Example of adding registered word / meaning number set to the meaning number / registered word list

FIG. 6 is a schematic configuration diagram of a semantic recognition system according to a second embodiment of the present invention;

FIG. 7 is a flowchart showing an overall processing flow of the semantic recognition system according to the second embodiment of the present invention;

FIG. 8 shows an example of an external description file 50.

FIG. 9 is an example in which the acquired meaning numbers are expanded to the meaning number / registered word list 41.

FIG. 10: Generated meaning number / registered word list 41
Example

FIG. 11 is a schematic configuration diagram of a semantic recognition system according to a third embodiment of the present invention;

FIG. 12 is a flowchart showing an overall processing flow of the semantic recognition system according to the third embodiment of the present invention;

FIG. 13 shows an example of a generated meaning number / registered word list.

FIG. 14 is a schematic configuration diagram of a semantic recognition system according to a fourth embodiment of the present invention.

FIG. 15 illustrates an example of a recording medium.

[Explanation of symbols]

Reference Signs List 10 voice input device 20 voice signal file 30 voice recognition unit 31 voice recognition engine 32 voice recognition word update unit 40 recognition word / meaning number conversion unit 41 meaning number / registered word list 42 meaning number / registered word list changing unit 43 acquisition of meaning number Unit 44 Meaning number / registered word list generation unit 45 Registered word notification unit 50 Application 60 External description file 70 Language specification unit 80 Application server 90 External description file server 100 Client 110 Network 121 Recording medium such as hard disk at connection destination 122 CD- Portable recording medium such as ROM and flexible disk 123 CD-ROM 124 Flexible disk 125 Computer 126 Recording medium such as RAM / hard disk on computer

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Kenji Yamamoto 4-1-1, Kamiodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa Prefecture Inside Fujitsu Limited (72) Inventor Masahide Noda 4-1-1, Kamiodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa Prefecture No. 1 Inside Fujitsu Limited

Claims (9)

[Claims] 1. A system for providing a user interface capable of voice input to a user and providing an application interface for transmitting an input content of the user to an application, wherein the input audio signal is received. A voice signal input unit, a voice recognition unit for recognizing a corresponding word based on the input voice signal, a meaning number representing the meaning of the word, and a meaning number / registered word list which is a list of registered words belonging to the meaning number A recognition word / meaning number conversion unit, an application interface, and an application that handles the meaning number as data, and a word corresponding to the speech recognition unit based on the speech signal input to the speech signal input unit. Recognize and convert the recognized word into the corresponding meaning number by the recognized word / meaning number conversion unit , Meaning recognition system and notifies the meaning number the converted via the interface application to application. 2. The meaning recognition system according to claim 1, further comprising a meaning number / registration word list updating unit for updating the meaning number / registration word list of the recognition word / meaning number conversion unit. 3. An external description file in which a word / semantic system table in which a systematic description of systematically classifying meanings and encoding of each semantic unit and registered words belonging to each semantic number is recorded. The semantic recognition system according to claim 1, comprising: 4. The application in which at least a part of the processing data includes a meaning number, wherein the recognition word / meaning number conversion unit converts the meaning described in the application via the application interface. A meaning number acquiring unit for acquiring a number, a meaning number for extracting a registered word corresponding to the acquired meaning number from the word / semantic system table of the external description file, and generating the meaning number / registered word list.
The semantic recognition system according to claim 3, further comprising a registered word list generation unit. 5. The application in which the processing data includes a meaning number in at least a part of the processing data, and based on the meaning number described in the application itself, a word / semantic table of the external description file is used. A processing step of extracting a corresponding registered word; and a processing step of notifying the meaning number and the registered word to the recognition word / meaning number conversion unit, wherein the recognition word / meaning number conversion unit obtains from the application. 4. The meaning recognition system according to claim 3, further comprising a meaning number / registered word list generation unit that generates the meaning number / registered word list based on the meaning number and the registered word. 6. The registered word / semantic number conversion unit includes a registered word notifying unit that notifies the registered words included in the generated meaning number / registered word list to the voice recognition unit.
The semantic recognition system according to claim 4, wherein the voice recognition unit includes a voice recognition word updating unit that updates the content of a voice recognition engine based on the notified registered word. 7. A use language designation unit for designating a language used by a user, an external description file for each of a plurality of languages as the external description file, and a speech for each of a plurality of languages as a speech recognition engine of the speech recognition unit. A recognition engine, wherein the semantic number / registered word list generating unit generates a semantic number / registered word list based on an external description file of the language specified by the working language specifying unit, and the speech recognition unit includes: 7. The meaning according to claim 6, wherein a multi-lingual compliant user interface is provided that allows a user to input an instruction in a plurality of languages by using a speech recognition engine of a language specified by the language-in-use specifying unit. Recognition system. 8. An externally described file server that provides the externally described file, an application server that provides the application, the voice signal input unit, the voice recognition unit, the recognized word / meaning number conversion unit, and the application server. The semantic recognition system according to claim 6, further comprising a semantic recognition system client having an interface, wherein the externally described file server, the application server, and the semantic recognition system client are connected via a computer network. 9. A program describing processing steps for providing a user interface capable of voice input to a user and realizing a system for providing an application interface for transmitting a user input to an application. A voice signal input processing step of receiving a voice signal to be input, a voice recognition processing step of recognizing a corresponding word based on the input voice signal, a meaning number representing a meaning, and a registration belonging to the meaning number. A meaning number / registered word list generating processing step of generating a meaning number / registered word list which is a list of words, and converting the word recognized by the voice recognition processing step to a meaning number corresponding to the generated meaning number / registered word list. Recognized words to be converted
A computer-readable recording medium recording a semantic recognition processing program comprising a semantic number conversion processing step and a semantic number notification processing step of notifying the converted meaning number to the application.